# Alternative active elliptical crossover

I realised that the elliptic filters given by FvM and presented on my original Gleanings page, could be scaled to crossover frequencies so here they are:

The originals were created on the Nuhertz program which is not available to me otherwise I would try and improve the performance even more! C1,3 and C4 are non standard values that could be made by paralleling 5.6nF with 680pf. Crossover frequency is 2.2kHz. The table shows how to get different frequencies

Frequency | R1-3 | C1,3,4 |

1.8k | 12k | 7.6n |

2k | 11k | 6.9n |

2.2k | 10k | 6.3n |

2.4k | 9.1k | 5.7n |

2.7k | 8.2k | 5.2n |

The capacitor values are 100pF more than standard E24 values. You could use standard C values and make the R1-3 resistors correspondingly smaller. Tolerances should be 1%. Multiply the C value by 1.588 to get the best R value.

Having scaled the design to crossover frequencies, I could not resist scaling the HPF to a rumble filter. It is -3dB at 21Hz but 20 dB down at 17.5Hz. I have come full circle on the equal component 3^{rd} order Sallen & Key filter, having first got interested in 2010 in “Xmax problems?” When I tried modelling this with more sophisticated op-amp models than the ideal op-amp shown, the circuit showed a tendency to oscillate at 4MHz, so I investigated further.

My normal op-amp for filter applications is the NE5532, however R8 is a high value and the TL072 would fit the bill better here. The LT1057 is the nearest standard model to the TL072 I could find. I used the standard fix for a capacitive load on the op-amp, a resistor R1, but I had to go up to 2.2k to reduce the tendency to oscillation (green trace on the waveform diagram). The LPF is not affected by this problem. I would breadboard a HPF before committing to a design.

## Leave a Reply